In recent years, high-throughput techniques have revealed considerable structural organization of the human genome with diverse regions of the chromatin interacting with each other in the form of loops. Some of these loops are quite complex and may encompass regions comprised of many interacting chain segments around a central locus. Popular techniques for extracting this information are chromatin interaction analysis by paired-end tag sequencing (ChIA-PET) and high-throughput chromosome conformation capture (Hi-C). Here, we introduce a physics-based method to predict the three-dimensional structure of chromatin from population-averaged ChIA-PET data. The approach uses experimentally-validated data from human B-lymphoblastoid cells to generate 2D meta-structures of chromatin using a dynamic programming algorithm that explores the chromatin free energy landscape. By generating both optimal and suboptimal meta-structures we can calculate both the free energy and additionally the relative thermodynamic probability. A 3D structure prediction program with applied restraints then can be used to generate the tertiary structures. The main advantage of this approach for population-averaged experimental data is that it provides a way to distinguish between the principal and the spurious contacts. The program source-code is available at https://github.com/plewczynski/looper.

中文翻译：

---

基于自由能的 CTCF 介导的人类基因组染色质循环模型

近年来，高通量技术揭示了人类基因组的相当大的结构组织，染色质的不同区域以环的形式相互作用。这些环中的一些非常复杂，可能包含由围绕中心位点的许多相互作用链段组成的区域。提取此信息的流行技术是通过配对末端标签测序 (ChIA-PET) 和高通量染色体构象捕获 (Hi-C) 进行的染色质相互作用分析。在这里，我们介绍了一种基于物理学的方法，从人口平均的 ChIA-PET 数据中预测染色质的三维结构。该方法使用来自人类 B 类淋巴母细胞的实验验证数据，使用探索染色质自由能景观的动态编程算法生成染色质的 2D 元结构。通过生成最优和次优元结构，我们可以计算自由能和相对热力学概率。然后可以使用具有应用约束的 3D 结构预测程序来生成三级结构。这种方法对于总体平均实验数据的主要优点是它提供了一种区分主要接触和虚假接触的方法。程序源代码可从 https://github.com/plewczynski/looper 获得。通过生成最优和次优元结构，我们可以计算自由能和相对热力学概率。然后可以使用具有应用约束的 3D 结构预测程序来生成三级结构。这种方法对于总体平均实验数据的主要优点是它提供了一种区分主要接触和虚假接触的方法。程序源代码可从 https://github.com/plewczynski/looper 获得。通过生成最优和次优元结构，我们可以计算自由能和相对热力学概率。然后可以使用具有应用约束的 3D 结构预测程序来生成三级结构。这种方法对于总体平均实验数据的主要优点是它提供了一种区分主要接触和虚假接触的方法。程序源代码可从 https://github.com/plewczynski/looper 获得。